tao-train-deformable-detr
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the generation of shell commands to run model training, evaluation, and deployment tasks within Docker containers using the NVIDIA TAO Toolkit. This is the primary function of the skill and aligns with the provided documentation.
- [INDIRECT_PROMPT_INJECTION]: The skill processes dataset annotation files (e.g., annotations.json), which is a standard attack surface for indirect prompt injection. The risk is minimized as the data is consumed by specialized computer vision tools rather than directly influencing high-privilege agent decisions.
- Ingestion points: Annotation files referenced in dataset.test_data_sources.json_file within SKILL.md and references/skill_info.yaml.
- Boundary markers: The instructions do not define specific delimiters for separating data from instructions within the JSON annotation files.
- Capability inventory: The skill allows the agent to execute shell commands and manage files in the workspace via Docker.
- Sanitization: The skill relies on the TAO Toolkit's internal parsing logic for COCO format validation and does not prescribe additional sanitization steps.
Audit Metadata